FlexE

What Is FlexE?

FlexE, also known as Flexible Ethernet, is an interface technology that enables the implementation of network slicing and service isolation within a bearer network. It introduces a more flexible approach by decoupling the mapping between the MAC (Media Access Control) and PHY (Physical) layers. This decoupling allows for precise management and allocation of interface resources, effectively addressing the challenge of accommodating varying service requirements and network capabilities from different customers. Additionally, FlexE caters to specific industry demands, such as providing hard pipe isolation and facilitating on-demand allocation of bandwidth resources.

The Significance of FlexE

As services and application scenarios become more diverse, the demand for higher bandwidth on bearer networks is increasing. Customers desire a unified network to transmit various services, including home broadband, private line access, and mobile bearer services. However, standard Ethernet interfaces face several challenges in meeting these requirements:

1. Lack of flexible bandwidth granularities: Standard Ethernet interfaces are limited by the predefined rate ladder (e.g., 10G, 25G, 40G, etc.) defined by standards, preventing them from providing more flexible bandwidth options to accommodate diverse services.

2. Synchronization issues between IP and optical transmission devices: The development of IP and optical transmission devices is not always synchronized. For instance, optical transmission devices may not support certain interface rates like 25G or 50G, which must match the Ethernet rate of IP devices when interconnected.

3. Insufficient support for enhanced QoS capabilities: Standard Ethernet interfaces prioritize packet scheduling based on QoS packet priorities, which can lead to long packets blocking the pipeline and increasing latency for short packets, causing mutual service interference.

FlexE addresses these challenges and offers the following benefits:

1. Flexible bandwidth granularities: FlexE enables the configuration of interface rates, providing more flexibility to meet the requirements of diverse services and applications.

2. Decoupling from optical transmission device capabilities: FlexE decouples the Ethernet interface rate of IP devices from the link rate of optical transmission devices. This allows for the maximum utilization of existing optical transmission networks (OTN) to support Ethernet interfaces with new bandwidths.

3. Enhanced QoS capabilities for multi-service bearing: FlexE provides channelized hardware isolation on physical-layer interfaces, ensuring Service Level Agreement (SLA) assurance and isolated bandwidth for different services.

FlexE's general architecture

FlexE operates by decoupling the MAC (Media Access Control) layer from the PHY (Physical) layer through the introduction of the FlexE shim layer. This architecture enables flexible rate matching and provides various functions for bandwidth configuration. FlexE consists of three main components: the FlexE client, FlexE shim, and FlexE group.

1. FlexE client: represents the user interface that behaves similarly to traditional service interfaces in IP/Ethernet networks. It supports different Ethernet MAC data streams with variable rates, allowing for flexible configuration to meet specific bandwidth requirements. The FlexE client sends encoded bit streams of 64B/66B to the FlexE shim layer.

2. FlexE shim: functions as an intermediary layer positioned between the MAC and PHY layers, facilitating the implementation of essential FlexE capabilities via calendar slot distribution.

3. FlexE group: comprises multiple Ethernet PHYs. By default, the PHY bandwidth is divided based on the 5G granularity.

Three Primary Functions Within its General Architecture：

1. Bonding: Multiple PHYs can be bonded together to support higher rates. For instance, four 100G PHYs can be combined to achieve a 400G MAC rate.

2. Channelization: Multiple low-rate MAC data streams can share one or more PHYs. This allows for efficient utilization of bandwidth. For example, multiple MAC data streams (25G, 35G, 20G, and 20G) can be carried over a single 100G PHY, or three MAC data streams (125G, 150G, and 25G) can be distributed across three 100G PHYs.

3. Sub-rating: With sub-rating, a single low-rate MAC data stream can share one or more PHYs. For instance, a 100G PHY can carry only 50G MAC data streams. Sub-rating can be seen as a subset of channelization.

Application Scenarios of FlexE

Converged transport is becoming an inevitable direction for IP networks as the communications industry faces new challenges in achieving fast on-demand networking and flexible resource configuration. To address these challenges and meet the requirements of emerging applications and services, FlexE offers solutions based on its bonding, channelization, and sub-rating functions. FlexE solutions are well-suited to underpin the future network architecture with a focus on service experience. These solutions effectively cater to the evolving demands of bandwidth-intensive services like video, VR/AR, and 5G services. At present, FlexE finds extensive application in scenarios involving ultra-high-bandwidth interface deployment, IP+optical flexible networking, and network slicing.

Ultra-High-Bandwidth Interface Deployment

FlexE utilizes bonding technology to create ultra-high bandwidth interfaces. By combining multiple physical interfaces, such as four 100G PHYs, FlexE achieves higher MAC rates, such as a 400G interface. This approach optimizes bandwidth utilization, resolves existing link limitations, and ensures efficient data flow distribution.

IP+Optical Flexible Networking

FlexE interfaces serve as User-Network Interfaces (UNIs) that connect routers to optical transmission devices. Through flexible rate matching, the bandwidth of data flows carried by the UNIs can correspond to the bandwidth of the optical transmission device interfaces. This simplifies interface mapping, reduces device complexity, and lowers capital and operating expenses. The FlexE standards offer different modes for interconnection, with the unaware mode being recommended for utilizing existing optical transmission devices without requiring hardware upgrades.

Network Slicing

Network slicing is a critical concept to manage and optimize network resources based on different service requirements. FlexE plays a crucial role in implementing network slicing by providing logical networks (slices) on a shared network infrastructure. Each slice caters to specific service types or industry users, offering differentiated logical topology, SLA requirements, reliability, and security levels. FlexE technology, when combined with routers, enables the configuration of multiple FlexE interfaces (logical interfaces) on physical interfaces, ensuring strict isolation of bandwidth resources between interfaces. This facilitates the creation of secure, reliable, and dedicated network slices for various services or industries.

Key characteristics of FlexE interface-based network slicing include:

• Good Slicing Performance: Stable latency, no packet loss, and strict isolation between slices, ensuring guaranteed bandwidth and non-interference between services.

• Fine-Grained Slicing: FlexE supports a minimum slicing granularity of 1G, allowing for precise allocation of resources.

• Multiple Slices: FlexE, in conjunction with other resource reservation technologies like channelized sub-interfaces or Flex-channel, supports hierarchical slicing to accommodate complex service isolation requirements.

• Fast Slicing: FlexE enables rapid deployment of slices, with deployment times as short as minutes. Slice resources can be pre-deployed through an intelligent network controller or allocated on-demand, facilitating quick service deployment.